Re: [Nautilus-list] Some RH patches.

[Owen Taylor <otaylor redhat com>]

Alex Larsson <alexl redhat com> writes:

> +	const nsString str (uniStr);
> +	sSize = str.Length ();
> +
> +	/* allocate a wide string big enough to hold the unicode string,
> +	 * this is necessary since wchar_t is 32-bits with glibc */
> +	wide = g_new0 (wchar_t, sSize + 1);
> +	for (i = 0; i < sSize + 1; i++) {
> +		wide[i] = uniStr[i];
> +	}
> +
> +	/* use glibc function to determine the size of this string once
> +	 * encoded to a locale specfic multibyte string */
> +	count = wcstombs (NULL, wide, 0);
> +
> +	/* check for success */
> +	if (count == -1) {
> +		/* let Mozilla do a (lossy) conversion then */
> +		nsCString str;
> +		str.AssignWithConversion (uniStr);
> +		g_free (wide);
> +		return g_strdup (str.get ()); /* FIXME strdup needed? */
> +					      /* Phil: Oh Yes! indeed. */
> +	}

This code is not portable since it assumes the wchar_t is ISO-10646.

You can surround it with:

#ifdef __STDC_ISO_10646__

But I'm not sure what you'd do as a fallback ... handle ascii only?

(Also there is a question here whether the Mozilla string is UCS-2,
or UTF-16 with possible surrogate pairs.)

Regards,
                                        Owen